Gatik is building B2B autonomous vehicles for the $120 Billion short-haul logistics market. Unlike automation of large commercial vehicles such as Class 8 trucks, Gatik opts for light commercial vehicles -- class 1 through class 4 vehicles (16,000 lbs and under).
As we have seen in other forms of autonomy, the selection of vehicle typically has a large bearing on the type of service you offer. Gatik's selection of these vehicles point the company toward automating commercial trips between distribution centers (aka 'DCs') and between DCs and storefronts; they do not automate trips to end users or create small delivery robots as has been the typical path for local logistics automation. Gatik aims build the best service for automating light commercial routes. They are not building a consumer-facing tool; their service is available only for enterprise clients like beverage companies, grocers and others who move large quantities of goods in fixed routes from DCs to smaller locations and back all day.
On the technical side, Gatik's approach is unique as given its focus on hyper-optimized deep learning models. Most teams that use deep learning build route agnostic general perception models. Gatik uses a series of modules of deep learning models overfitted and hyper-optimized to specific routes across their entire autonomy stack, and applies these modular models via a deterministic rule based stack with rich priors. The team believes true autonomy can only be achieved by pairing deep learning with conventional robotics techniques (rules) to provide redundancy. They call this their 'Explainable AI,' and it allows Gatik to build safe, robust autonomous systems with exponentially less data. In short, Gatik blends machine learning with a more formal methods approach in all areas of their system.
Gatik’s team is composed of exceptional talent across autonomy and logistics. CEO Gautam Narang and CTO Arjun Narang come from Carnegie Mellon’s Robotics Institute, while Chief Engineer Apeksha Kumavat was the former perception lead at Ford’s Self-Driving team. The remaining engineering and operations team is comprised of former executives and directors from Uber ATG, Starsky, Pony.ai, Cruise, Becker Logistics, and Ryder Systems.
The core risk for Gatik is not technical -- although risks certainly remain in that area. The largest risk for Gatik is its sales channel. Gatik is aiming to put vehicles on the road in service of enterprise clients and that will require a dedicated sales effort with large, incumbent buyers. It is likely to be a long sales cycle with long periods of indecision by Fortune 1000 customers.
Another key risk for the company will be regulatory environments for commercial automation which are still largely undefined. NHTSA’s proposed guidelines for vehicle automation currently focus on passenger robotaxi applications and commercial businesses have mostly been left out of current language. Gatik must partner with other commercial automation startups to create a uniform set of national frameworks for commercial vehicle automation.
We look forward to continuing our support for the Gatik team as they grow and build leadership in light commercial vehicle transport.